JP2002374417A - Device and method for image processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a plurality of photographing devices which can output the same images even when the photographing devices take pictures by making those devices have the same gradation curve. SOLUTION: An image processor has an image information acquiring means (200) which acquires image information showing the relation between pixel values and density (or luminance, etc.), and a gradation curve deriving means which derives a gradation curve from the image information. Further, the processor is equipped with an error decision means (220) which decides an error between the image information and the gradation curve to derive a precise gradation curve for the image information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus and an image processing method, and is particularly suitable for acquiring and displaying an image of a subject and automatically adjusting gradation processing of the image.

[0002]

2. Description of the Related Art A medical X-ray radiation image has a long history of converting an X-ray intensity distribution transmitted through a human body into a fluorescent distribution of a fluorescent substance, directly transferring the light intensity to a silver halide film, and developing the image. In recent years, a radiation image is read out by various methods such as a method of reading a stimulable phosphor as a latent image of an X-ray intensity distribution as energy, a method of directly reading out a fluorescent distribution of the phosphor by X-rays as an image, and a technique not using fluorescence. 2. Description of the Related Art Digital images are being constructed by reading them out as electric signals and performing digital conversion.

[0003] By using digital images in the same way as CT and other devices, the efficiency of filing, the practical use of remote diagnosis, and the improvement of medical treatment technology and efficiency can be improved. Furthermore, various image processings have become possible, and diagnostic methods are changing. Among them, when outputting an obtained digital image to a silver halide film or when displaying image information on an image display device such as a CRT, it is necessary to interactively perform a gradation process for capturing a region of interest of the image and making it easy for a doctor to diagnose. And output.

Japanese Patent Publication No. Sho 60-42964 discloses an example of gradation processing using a nonlinear curve. The gradation processing using such a non-linear curve effectively compresses a portion of the image information other than the desired region of interest, and reduces the gradation range included in the input information without substantially reducing the information amount of the region of interest. It has features that can be displayed efficiently. As an example of a non-linear curve in Japanese Patent Publication No. 60-42964, the following curve equation is used in the embodiment.

[0005]

(Equation 5)

Here, the pixel value x proportional to the X-ray dose and D
₀, D_maxIs the minimum density, the maximum density, D _m= (D_max-D₀)
/ 2 is the central density and a is the scale factor.

This equation has the following features. First, for the sake of explanation, the variables are redefined as follows. The pixel value which is Log with respect to the X-ray dose is X = logx, and the translation amount is B =
logx _1, D _min and D _max minimum density and maximum density, and the A is the scale factor. Then, the above equation can be described as follows.

[0008]

(Equation 6)

FIGS. 16 and 17 show the role of the parameters A and B of the gradation curve. Parameter A adjusts the slope of the gradation curve, and Parameter B adjusts the parallel movement of the gradation curve.

[0010]

In general, the adjustment of the gradation processing is a problem that the image is output after adjusting the parameters, and if the user does not like the adjustment, the image is output after adjusting again. There is.

Further, there is a problem that it is not possible to express a gradation curve corresponding to various images with one determined function. Further, there is a problem that a desired gradation process cannot be obtained only by a gradation curve provided in advance in the apparatus.

When a new digital photographing apparatus is introduced, there is a demand for comparison with a past analog film image or an image photographed by an old photographing apparatus.
When diagnosing an image photographed at a different facility by remote diagnosis or the like, there is a demand to adjust the gradation of the image photographed by a familiar photographing apparatus so that the gradation is obtained.

The present invention has been made in view of the above-mentioned problems, and when adjusting a gradation process having a desired gradation characteristic, an operator can easily adjust the gradation processing and can adjust a gradation characteristic of an image output by another photographing apparatus. It is an object of the present invention to provide a method of deriving a gradation curve having the same gradation characteristics as above and performing gradation processing on an image, and an image processing apparatus and method having a display device using the same.

[0014]

According to one aspect of the present invention, there is provided image information acquiring means for acquiring image information representing a relationship between a pixel value and an output value corresponding thereto, and a tone curve is derived from the image information. And a gradation curve deriving means, which is a linear sum of the functions F _i (X−b _i , a _i ), and C _i , a _i , b _i , D _min , D There is provided an image processing apparatus characterized in that a model function represented by Expression 1 is used with _max as a parameter.

According to another aspect of the present invention, an image information obtaining step of obtaining image information representing the relationship between the output value and the corresponding pixel value, the function _{F i (X-b i,} a i) linear A tone curve deriving step of deriving a tone curve from the image information using a model function represented by the following equation 1 with the sum being C _i , a _i , b _i , D _min , and D _max. An image processing method characterized by having the following.

According to the present invention, the gradation curve is represented by a function f _i (X
−b _i , a _i ), a gradation curve with a different slope can be created in a different pixel value area, and a gradation curve suited to various images can be created.

[0017]

(First Embodiment) A first embodiment of the present invention will be described below. FIG. 1 is a diagram showing a configuration of an imaging device according to an embodiment of the present invention. Radiation is radiated from 120 sources to 120 subjects, and 120
The radiation image of the subject enters the sensor 130. The radiation image is acquired as a digital image (hereinafter, image) by the A / D converter 140. The operation unit 150 extracts a characteristic region of the image from the acquired image of the subject, and outputs an output value (specifically, an output value (specifically, Density, luminance, etc., which will be referred to as density hereinafter.) The (pixel value, density) pair information obtained by the input is sent to the image processing unit 160 together with the image, subjected to gradation processing, and output to the image display unit 170 such as a monitor or a printer.

FIG. 2 shows the flow of processing in the operation unit 150 and the image processing unit 160. The operation unit input information is obtained by 200 image information obtaining means. The operation unit input information is several pieces of (pixel value, density) pair information.
A tone curve is derived from these (pixel value, density) by a tone curve deriving means 210. Next, the error between the input (pixel value, density) and the gradation curve is evaluated by the error determination means 220, and if it is not good, the process returns to the gradation curve derivation means 210, and the type of the gradation curve is changed. A gradation curve having a good match with (pixel value, density) is derived. If a tone curve having a good match with (pixel value, density) is obtained by the error determining means 220, the subject image is subjected to tone processing by the tone processing means 230. The processed image is sent to the display unit and displayed.

Hereinafter, the flow of the processing in FIG. 2 will be described in detail. First, 200 image information acquisition units will be described. The image of the subject 120 shown in FIG. 1 is temporarily displayed on an image display device in the operation unit 150. The operator selects a region of interest in the image while viewing the image, and obtains a pixel value by taking an average value of the region. Further, by inputting the densities corresponding to the selected region of interest, several pairs of (pixel values, densities) are obtained.

A more detailed description of the 200 image information acquisition means will be described later. Next, the derivation of the tone curve by the tone curve deriving means 210 and the error determining means 220 from this pair information of (pixel value, density) Will be described.

The pair information of (pixel value, density) is expressed as (X, Y) =
(Pixel value, density), which is parameter-fitted by the following model function.

[0022]

(Equation 7)

However,

[0024]

(Equation 8)

## EQU1 ##

Here, D _min and D _max are the minimum density (minimum luminance) and the maximum density (maximum luminance) of the image. C _i is the number of positive 1 below. N is a natural number, and is increased by one from N = 1 in the 220 error determination means.

The function of F _i (X−b _i , a _i ) has the following properties.

[0028]

(Equation 9)

Here, θ (X) is θ (X) when X> 0.
= 1 and X <0 is a step function that satisfies θ (X) = 0.

As a specific example of such f _i (X, a _i ), there is a function shown in Table 2 below. Since Table 2 is the same as Table 1, it is referred to as Table 1 below.

[0031]

[Table 2]

Here, Slop (X, a _i ) has the following properties.

[0033]

(Equation 10)

The properties of the gradation curve according to the model function of Equation 1 will be described below. First, the function f in Equation 1
_{i (X-b i, a} i) describes the nature of.

FIG. 6 is a graph of the function shown in Table 1.
Here, the function of Table 1 is defined as f _i (X−b _i , a _i ) and X
Axially by b _i are translated. As can be seen from this graph, it can be seen that the functions in Table 1 have the properties of Equations 3 and 4.

FIG. 7 shows the first function in Table 1.

[0037]

[Equation 11]

In the above, the value of a _i gradually approaches 0. As can be seen from this graph, a _i
Is gradually approached to 0, the function of Table 1 has the property of Equation 5. B _i as described above represents the center of the gradient of the function, a _i it is found that indicates the degree of inclination.

Next, in the equation 1,

[0040]

(Equation 12)

The nature of the part will be described.

FIG. 8 shows the first table in Table 1.

[0043]

(Equation 13)

Linear sum of

[0045]

[Equation 14]

4 is a graph showing Each f _i (X−
b _i , a _i ) is a function that monotonically increases from 0 to 1 as shown in FIG. Linear sum of those functions

[0047]

(Equation 15)

The equation (2)

[0049]

(Equation 16)

In the same manner, the function becomes a function that monotonically increases from 0 to 1. However, Slop (X,
In the case of _ai ), the parallel movement amount b _i is limited so as to increase monotonically. Figure 8 function

[0051]

[Equation 17]

C _i indicates the component ratio of f _i (X−b _i , a _i ) having the degree of inclination a _i and the amount of parallel movement b _i .

Therefore, as shown in FIG.

[0054]

(Equation 18)

Is the degree of inclination a _{i in} each of the three pixel areas.
Can be expressed. Similarly,

[0056]

[Equation 19]

Is the degree of inclination a _{i in} each of N pixel areas.
Can be expressed.

[0058]

(Equation 20)

In the above, the function f _i (X−b _i , a _i ) may be a combination of the three functions shown in Table 1.

Next, a fitting method will be described. Since the operator specifies the region of interest and its density while viewing the image, the score of (X, Y) = (pixel value, density) is not very large. In such a case, it is particularly convenient to reduce the parameters to be fitted in the equation 1 as much as possible, since the gradation curve can be uniquely determined.

(1) When the information of (X, Y) = (pixel value, density) is one point In this case, in Expression 1, D _min , D _max , a _i , b _i , and C _i are fixed to certain optimum values. I do. In Expression 1, the value of N (corresponding to the type of the gradation curve) is selected in advance by the operator. By substituting XB into Y (X) in Equation 1, a new parameter B is introduced by setting it to Y (XB), and this B is adjusted so that the gradation curve passes through the point p. . The fitting method includes the Newton method. This example is shown in FIG. Here, the error determination means 220 is not executed after the fitting.

(2) When the information of (X, Y) = (pixel value, density) is two points At this time, in Expression 1, D _min , D _max , a _i , b _i , and C _i are fixed to certain optimum values. I do. In Expression 1, the value of N (corresponding to the type of the gradation curve) is selected in advance by the operator. Substituting (X−B) / A for Y (X) in equation 1, Y ((X−B) /
A), new parameters A and B are introduced, and only these two parameters A and B are adjusted to obtain 2
The point is fitted to (X, Y) = (pixel value, density). As a fitting method, first, as shown in FIG. 10, of (X, Y) = (pixel value, density) of two points,
B is adjusted so that the gradation curve passes through a point p1 close to the density of ( _Dmax + _Dmin ) / 2. Next, A is adjusted to match another point p2. In the process of adjusting A, the gradation curve changes to the point p.
If it deviates from 1, B is adjusted each time so that the gradation curve passes through p1. Here, the error determination means 220 is not executed after the fitting.

(3) When (X, Y) = (Pixel Value, Density) Information is Three or More Points At this time, in Expression 1, D _min , D _max , a _i , b _i , and C _i are set to certain optimum values. Fix it. Substituting (X−B) / A for Y (X) in Equation 1 and introducing new parameters A and B by making Y ((X−B) / A), these two parameters A and B Is adjusted, and two points (X, Y) = (pixel value,
Concentration).

As a fitting method, as shown in FIG. 10, among the three or more points (X, Y) = (pixel value, density), the density closest to the density (D _max + D _min ) / 2 is obtained. B is adjusted so that the gradation curve passes through the point p. Next, A is adjusted so that the sum of the distances between the other points and the gradation curve (the sum of the squares of the distances is also good) is minimized. If the tone curve deviates from point p in the process of adjusting A, B is adjusted each time
So that the gradation curve passes.

As another method, if the number of points (X, Y) = (pixel value, density) is large, D _min and D _max are not fixed, and A, B, D _min and D _max are used as parameters. , Leve
There is also a fitting method using the nberg-Marquardt method or the like.

After the fitting, the error judgment means 220 judges the error between (X, Y) = (pixel value, density) and the gradation curve. Returning to the tone curve deriving means 210, the value of N in Equation 1 is changed (corresponding to changing the type of the tone curve).

Similarly, fitting and error judgment are performed, and
Stop by fitting at the value of N where the error is within a certain threshold. All N values (all gradation curve types)
If the error does not fall within a certain threshold value, the gradation curve of the value of N having the smallest error is selected. As the evaluation function used to determine the error, for example, the sum of squares of the difference between the density at the (pixel value, density) point and the density Y in Expression 1 at the corresponding pixel value can be used.

As described below in (1) to (3),
Substituting (X−B) / A and X−B for Y (X) in equation 1, Y
The advantage of introducing new parameters A and B by setting ((X−B) / A) and Y (X−B) will be described.

In FIG. 8, equation 1 is_min, D_max, A_i,
b_i, C_iBy changing the value of each, various types
(Tilt, etc.) was obtained. Only
Then D_{mi n}, D_max, A_i, B_i, C_iPlace to directly change
Is too many parameters to change,
When the score of (X, Y) = (pixel value, density) is small,
The parameter values do not converge well. (X)
-B) / A and XB are substituted, and parameters A and B are changed.
To be converted into a_i, B_i, C_iIndirect value of
It corresponds to changing it. It can be seen from Equation 1 and FIG.
Thus, parameters A and B are the slopes of the gradation curve, respectively.
And indirectly adjust the degree of translation. This
(X, Y) = (pixel value, density)
In this case, (X−B) / A and X−B are substituted for Y (X),
By indirectly adjusting parameters A and B,
Parameter D_min, D_max, A_i, B_i, C_iConverges well
It is possible to make it.

After the derivation of the tone curve by the tone curve deriving means 210 and the error determining means 220, the tone processing means 230 will be described.

In FIG. 2, the subject image is input to the gradation processing means 230 and the gradation curve deriving means 210 and 22
0 based on the gradation curve by the error determination means of 0
Is converted into an output pixel value corresponding to a digital value such as a luminance if the input image is a monitor or a density if the printer is a printer, and the image is displayed or output.

FIG. 11 shows a method of creating an LUT. First,
The conversion rule between the input pixel value and the relative pixel value, the conversion rule between the relative pixel value and the density and brightness based on the derived gradation curve, and the conversion rule between the density and brightness and the output pixel value depending on the monitor or printer etc. Derives a conversion rule between the input pixel value and the output pixel value, and this becomes the LUT.

The parameter storing means of 240 is 210
Of the parameters A, B, and N (corresponding to the type of the tone curve) derived by the tone curve deriving means and the error determining means 220 and the specific function used (described in Table 1)
Is stored, and if there is a change in the conversion rule between the output pixel value and the density and luminance depending on the monitor or the printer, the LUT is used again to derive the LUT.

Based on the above description, the 200 image information obtaining means will be described in more detail. The above description is based on the premise that information of (X, Y) = (pixel value, density) has been obtained from the operator. It is easier for the operator to obtain the tone curve desired by the operator if the operator determines the tone curve while actually looking at the image. Then, the operator sets (X, Y) = (pixel value,
Once the information is input, the image is output once, and if the operator does not like the image subjected to the gradation processing, (X,
Y) = (pixel value, density) information is input.
For this reason, as shown in FIG. 12, the image information obtaining means 200 has a part for displaying an image and a part for adjusting a parameter integrated with each other. The image display of FIG. 2 is integrated with 200 image information acquisition means. Referring to FIG. When the operator selects a region of interest in the image while viewing the image, and inputs the density (or luminance or the like) of the region of interest, the current gradation curve is displayed on the coordinates of the (pixel value, density) displayed together with the image. And the newly selected point of (X, Y) = (pixel value, density) is displayed, which means that information of (X, Y) = (pixel value, density) has been obtained. Thereafter, parameter fitting is performed as described above, and the image is displayed again. This corresponds to the route U in FIG.

The 200 image information acquiring means includes buttons for finely adjusting parameters A, B, D _min , D _max , and N. By pressing a button or directly inputting a numerical value, the parameters of the gradation curve can be finely adjusted. This corresponds to the route T in FIG.

(Second Embodiment) Hereinafter, a second embodiment of the present invention will be described.
An embodiment will be described. FIG. 1 is a diagram showing a configuration of an imaging device according to an embodiment of the present invention. Radiation is emitted from the source 110 to the object 120, and a radiation image of the object 120 is incident on the sensor 130. The radiation image is acquired as a digital image (hereinafter, image) by the A / D converter 140. The operation unit 150 extracts a characteristic region of the image from the acquired image of the subject, and outputs an output value (specifically, an output value (specifically, Density, luminance, etc., which will be referred to as density hereinafter.) The (pixel value, density) pair information obtained by the input is sent to the image processing unit 160 together with the image, subjected to gradation processing, and output to an image display unit such as a monitor or a printer.

FIG. 2 shows the flow of processing in the operation unit 150 and the image processing unit 160. The operation unit input information is obtained by 200 image information obtaining means. The operation unit input information is several pieces of (pixel value, density) pair information.
A tone curve is derived from these (pixel value, density) by a tone curve deriving means 210. Next, the error between the input (pixel value, density) and the gradation curve is evaluated by the error determination means 220, and if it is not good, the process returns to the gradation curve derivation means 210, and the type of the gradation curve is changed. A gradation curve having a good match with (pixel value, density) is derived. If a tone curve having a good match with (pixel value, density) is obtained by the error determining means 220, the subject image is subjected to tone processing by the tone processing means 230. The processed image is sent to the display unit and displayed.

Hereinafter, the flow of the processing of FIG. 2 will be described in detail. First, 200 image information acquisition units will be described. When the subject 120 in FIG. 1 is, for example, a radiation attenuating plate whose thickness changes stepwise, a radiation image obtained as an image is an image as shown in FIG. A region corresponding to each step is extracted, and a pixel value is obtained by taking an average value of the region. The extraction method depends on whether the operator selects an area or
It is extracted by a binarization process. When the operator inputs the density corresponding to each of the extracted regions, some (pixel value, density) pair information can be obtained. This (pixel value, density)
Is plotted on a pixel value (relative value) -density coordinate as shown in FIG.

A more detailed description of the image information acquisition means 200 will be given later, and then the derivation of the tone curve by means of the tone curve deriving means 210 and the error judging means 220 from this pair information of (pixel value, density). Will be described. The pair information of (pixel value, density) is expressed as (X, Y) = (pixel value, density), and this is parameter-fitted by the following model function.

[0080]

(Equation 21)

However,

[0082]

(Equation 22)

Is as follows.

Here, D _min and D _max are the minimum density (minimum luminance) and the maximum density (maximum luminance) of the image. C _i is 1
The following positive number. N is a natural number, and is increased by one from N = 1 in the 220 error determination means.

The function of f _i (X−b _i , a _i ) has the following properties.

[0086]

(Equation 23)

Here, θ (X) is θ when X> 0.
It is a step function where θ (X) = 0 when (X) = 1 and X <0.

A specific example of such f _i (X, a _i ) is a function shown in Table 3 below. Table 3 is the same as Table 2 and will be referred to as Table 1 below.

[0089]

[Table 3]

Here, Slop (X, a _i ) has the following properties.

[0091]

(Equation 24)

As a fitting method, Levenberg-
Marquardt method and the like. FIG. 5 shows Equation 1.

[0093]

(Equation 25)

This is an example of fitting in the case where.

The properties of the gradation curve according to the model function of Equation 1 will be described below. First, the function f in Equation 1
_{i (X-b i, a} i) describes the nature of.

FIG. 6 is a graph of the function shown in Table 1.
Here, the function of Table 1 is defined as f _i (X−b _i , a _i ) and X
In the axial direction by b _i are translated. As can be seen from this graph, it can be seen that the functions in Table 1 have the properties of Equations 3 and 4. Figure 7 shows the first function in Table 1.

[0097]

(Equation 26)

In the above, the value of a _i gradually approaches 0. As can be seen from this graph, a _i
Is gradually approached to 0, the function of Table 1 has the property of Equation 5. B _i as described above represents the center of the gradient of the function, a _i it is found that indicates the degree of inclination.

Next, in the equation 1,

[0100]

[Equation 27]

The nature of the portion will be described.

FIG. 8 shows the first table in Table 1.

[0103]

[Equation 28]

Linear sum of

[0105]

(Equation 29)

FIG. Each f _i (X−
b _i , a _i ) is a function that monotonically increases from 0 to 1 as shown in FIG. Linear sum of those functions

[0107]

[Equation 30]

The equation (2)

[0109]

(Equation 31)

In the same manner, the function becomes a function that monotonically increases from 0 to 1. However, Slop (X,
In the case of _ai ), the parallel movement amount b _i is limited so as to increase monotonically. Figure 8 function

[0111]

(Equation 32)

C _i of [0112] I _{am, f i (X-b i} , a i) with a degree a _i and parallel movement amount b _i of each slope represents the component ratio of.

Therefore, as shown in FIG.

[0114]

[Equation 33]

Is the degree of inclination a _{i in} each of the three pixel areas.
Can be expressed. Similarly,

[0116]

(Equation 34)

Is the degree of inclination a _{i in} each of N pixel areas.
Can be expressed.

[0118]

(Equation 35)

In the above, the function f _i (X−b _i , a _i ) may be a combination of the three functions shown in Table 1.

Hereinafter, a method of fitting parameters will be described.

[0121] (4) (X, Y) = ( pixel value, density) at this time if a large number than the number of information parameters, firstly as N = 1 in Equation 1, is fixed to a certain value C ₁ (C ₁ = 1 from equation 2), D _min , D _max , a ₁ , b
The model function of Equation 1 is fitted with ₁ as a parameter. Thereafter, the error determination means 220 (X,
Y) = (pixel value, density) and an error determination between the gradation curve are performed, and if the error is equal to or greater than a certain threshold value, N is set to N = 2 in equation (1). At this time, D _min , D _max , b ₁ , b ₂ , C ₁ , a ₁ , a
The model function of Equation 1 is fitted with ₂ as a parameter. The value of C ₂ is determined by Equation 2. After this 2
(X, Y) = (pixel value, density)
Is determined, and if the error is equal to or greater than a certain threshold, then N = 3 in equation (1). At this time, D _min , D
_max, and the _{b 1, b 2, b 3} , C 1, C 2, a 1, a 2, a 3 to parameters, fitting the model function of formula 1. The value of C ₃ is determined by Equation 2. Thereafter, an error determination unit 220 determines an error between (X, Y) = (pixel value, density) and the gradation curve.
In Expression 1, N = i (i ≧ 4) is set.

At this time, D _min , D _max , b ₁ ,..., B _i ,
_{C 1, ..., C i-} 1, a 1, ..., and the a _i to the parameter,
The model function of Equation 1 is fitted. The value of C _i is determined by Equation 2. After this, the error determination means of 220
An error between (X, Y) = (pixel value, density) and the gradation curve is determined, and if the error is equal to or greater than a certain threshold, N = i + 1 is increased. If the error in this process is above a certain threshold, N
Is determined, and for all N, 22
An error determination means of (X, Y) = (pixel value, density) and a gradation curve are determined by an error determination means of 0. If the error is equal to or more than a certain threshold, (X, Y) = N A tone curve having an N value with a small error between (pixel value, density) and the tone curve is selected, and a tone curve is derived. As the evaluation function used to determine the error, for example, the sum of squares of the difference between the density at the (pixel value, density) point and the density Y in Expression 1 at the corresponding pixel value can be used.

After the derivation of the tone curve by the tone curve deriving means 210 and the error determining means 220, the tone processing means 230 will be described.

In FIG. 2, the subject image is input to the tone processing means 230, and the tone curve deriving means 210
0 based on the gradation curve by the error determination means of 0
Is converted into an output pixel value corresponding to a digital value such as a luminance if the input image is a monitor or a density if the printer is a printer, and the image is displayed or output. Figure 1 shows how to create an LUT
It is shown in FIG. First, the conversion rule between the input pixel value and the relative pixel value,
The conversion rule between the input pixel value and the output pixel value is based on the conversion rule between the relative pixel value and the density and brightness based on the derived gradation curve, and the conversion rule between the density and brightness and the output pixel value depending on the monitor or the printer. Is derived, and this becomes the LUT. Here, the input pixel value may be a value obtained by log-converting the X-ray dose,
It may be a proportional value and is not particularly limited.

The parameter storage means 240 is 210
The gradation curve derivation means and 220 parameters D _min, which is derived by the error determining _{means, D max, b 1, ...} , b i, C
_{1, ..., C i-1} , a 1, ..., ( corresponding to the type of gradation curve) values of a _i and N and specific functions that are used (as shown in Table 1)
Is stored, and if there is a change in the conversion rule between the output pixel value and the density and luminance depending on the monitor or the printer, the LUT is used again to derive the LUT.

Based on the above description, the 200 image information obtaining means will be described in more detail. Up to this point, an image in which the pixel value changes stepwise in each region is acquired by using, for example, a radiation attenuating plate or the like in which the subject 120 in FIG. , Etc.), the information of (X, Y) = (pixel value, density) is obtained, the gradation curve is determined, and the LUT is derived. It is better for the operator to determine the gradation curve while actually looking at the image.
It is easy to obtain a gradation curve required by the operator. Therefore, it is preferable that the image information acquisition unit 200 has a unit for displaying an image and a unit for inputting density (or luminance or the like) integrated as shown in FIG. The image display of FIG. 2 is integrated with 200 image information acquisition means.

Next, FIG. 14 will be described. When the operator selects each area of the image while viewing the image and inputs the density (or luminance, etc.) of each area, the current gradation curve is displayed on the coordinates of the (pixel value, density) displayed together with the image. And the newly selected point of (X, Y) = (pixel value, density) is displayed, which means that information of (X, Y) = (pixel value, density) has been obtained. Thereafter, parameter fitting is performed as described above, and the image is subjected to gradation processing again and displayed.

Further, in order to finely adjust the tone curve obtained by the parameter fitting, the parameters A, B, D _min ,
There are buttons for changing the values of D _max and N. The parameters A and B are obtained by substituting (X−B) / A for Y (X) in Equation 1,
This is a new parameter introduced by Y ((X−B) / A). These are parameters for adjusting the slope and B for the gradation curve. By pressing these buttons, the numerical value can be changed or the numerical value can be directly input, whereby the parameters of the gradation curve can be finely adjusted. This corresponds to the route T in FIG.

FIG. 13 is an example used when it is desired to greatly change the shape of the gradation curve once obtained for an image other than an image of a radiation attenuating plate or the like whose thickness changes stepwise. It is. Many (X, Y) =
There are (pixel value, density) points, and the positions of these points can be changed. However, since the gradation curve must be monotonically increasing, there is a restriction that the positions of these points are also changed so as to increase monotonically. The position of the point of the gradation curve is moved with a mouse or the like, and parameter fitting is performed in the same manner as described in the example of the attenuation plate. This corresponds to the route U in FIG. Then, the values of the parameters A, B, _Dmin , _Dmax , and N are finely adjusted while viewing the image. this is,
This corresponds to the route T in FIG.

The handling of the parameters A and B with respect to the gradation curve will be described in more detail. In Expression 1, Y (X) is expressed as (XB)
Substituting / A gives the following equation.

[0131]

[Equation 36]

Here, X corresponds to the relative pixel value, and Y corresponds to the density (luminance, etc.). By parameter fitting,
The parameters D _min , D _max , b ₁ , ..., b _N , C ₁ , ...,
_{C N-1, a 1,} ..., a N are determined (a value of C _N is determined from the equation 2.), The image subtle with fewer parameters is performed a fine adjustment of the after this parameter A, and B Gradation processing can be easily performed. Changing the values of the parameters A and B is indirectly based on a ₁ ,..., A _N ,
Since this corresponds to changing the values of b ₁ ,..., b _N , FIG.
As described above, the inclination and the parallel movement (parallel movement corresponds to changing the density (luminance and the like) of a certain relative pixel value) are changed. However, when changing the gradient of the gradation curve, it is easier to adjust the gradation process required by the operator by changing the gradient while keeping the density (luminance and the like) of the region of interest in the image constant. To accomplish this by substituting b _i -Dp the parameters b _i of Equation 9, to introduce a new parameter Dp.

[0133]

(37)

Hereinafter, description will be made with reference to FIG. The operator directly selects a density (luminance or the like) to be kept constant or indirectly selects a density by designating a region of interest from an image. A pair of the relative pixel value B and the selected density (such as luminance) is a fixed density point. At this time, the gradation curve has not yet passed through the fixed density point, and the value of the parameter Dp is changed so that the gradation curve matches the fixed density point. If the parameter A is changed when the gradation curve of Expression 10 passes through the fixed density point, the gradation curve changes around the fixed density point. As a method of fitting the value of the parameter Dp so that the gradation curve matches the fixed density point, there is a Newton method or the like.

As described above, the image processing apparatuses according to the first and second embodiments use the pixel value and the density (or luminance or the like).
Image information acquiring means for acquiring image information representing the relationship of (1) and a tone curve deriving means for deriving a tone curve from the image information. Further, an error determination unit for determining an error between the image information and the gradation curve is provided to derive an accurate gradation curve for the image information.

The embodiments of the present invention are applicable to the following various embodiments.

(1) Function F_i(Xb_i, A_iLinear sum of)
And C_i, A_i, B_i, D_min, D_{m ax}The parameter
Is characterized by using a model function represented by Equation 1 as
Gradation curve generator.

(2) The function F _i (X−b _i , a _i ) is a monotonically increasing function except for the case of a straight line (the function of 3 in Table 1), and has the properties of Equations 3 and 4. 1 of the coefficient C _i is characterized by satisfying the formula 2 (1) gradation curve generator according.

(3) The function F _i (X−b _i , a _i ) is
The gradation curve generator according to (2), wherein the function is one of the three functions shown in Table 1.

(4) The linearity of the function F _i (X−b _i , a _i ) is based on the pair information of the pixel value and the output value (density or luminance, etc.) in one-to-one correspondence (pixel value, density). Using the model function expressed by the expression 1 which is the sum, C _{i of} the expression 1,
_{a i, b i, D min} , gradation curve generating apparatus characterized by having a gradation curve derivation means for fitting the D _max as a parameter.

(5) The tone curve deriving means associates the derived tone curve with a pixel value and an output value (such as density or brightness) in a one-to-one correspondence (error between pixel information and density information). And a tone curve deriving means for increasing the value of the natural number N in Equation 1 and deriving the tone curve again if the error is equal to or greater than a certain threshold value. 4) The gradation curve generator according to the above.

(6) The parameters C _i , a _i , b _i , D _min , and D _max are fixed by the gradation curve deriving means, and the derived gradation curve Y (X) of the equation 1 is given by (X− B) / A
By introducing Y ((X−B) / A), new parameters A and B are introduced, and a pixel value selected on the image and an output value (such as density or luminance) are associated in a one-to-one correspondence (pixel value, The tone curve generating device according to (4), further comprising a tone curve deriving unit that adjusts parameters A and B to the pair information of (density) to perform fitting.

(7) The pair information of the pixel value for fitting the gradation curve and the output value (density or luminance, etc.) in one-to-one correspondence (pixel value, density) indicates the object whose density changes stepwise. (4) The image processing apparatus according to (4), further comprising an image information acquisition unit configured to capture the image data and input a density (or luminance or the like) with respect to a pixel value of a corresponding area of the displayed subject image. Generator.

(8) The gradation curve generator according to (4), wherein the fitting result of the gradation curve can be confirmed by displaying both the gradation processed image and the gradation curve.

(9) The gradation curve generating apparatus according to (8), wherein a designated (pixel value, density) point is plotted on the image displayed on the gradation curve.

(10) Pixel values for fitting the gradation curve described in (4) and output values (such as density or luminance) are associated in a one-to-one correspondence (pixel value, density)
Is registered for each gradation curve for which parameters are determined, and the registered (pixel value, density) points are plotted on the gradation curve described in (8).
By moving this plot point, the value of the (pixel value, density) point is changed, and the parameter of the gradation curve described in (1) is fitted to the new pair information of (pixel value, density). A gradation curve generator characterized by the following.

[0147] (11) above (1) by substituting b _i -Dp the parameters b _i of formula 1 in the gradation curve according to the new parameter Dp introducing the expression 10, the parameter Dp of formula 10 A tone curve generation device characterized by changing the value to change the density of a fixed density point that is the center of changing the slope of the tone curve.

According to the above (1), the gradation curve is converted to the function f _i.
The linear sum of (X−b _i , a _i ) has the effect that tone curves with different slopes can be created in different pixel value areas, and tone curves suited to various images can be created.

According to the above (2), the function f_i(X
-B_i, A_i) Satisfy the conditions of Equations 2 to 4 in the linear sum of
The lower the darkness, which is a parameter of the tone curve equation 1,
Degree D _minAnd the highest concentration D_maxWhen the image is fixed, the image
Gradation processing can be performed with the low density and the maximum density fixed.
This has the effect of making it easy for the operator to adjust the curve.

According to the above (3), the function f _i (X−b _i ,
Using the functions of Table 1 as a specific form of a _i ) can produce a tone curve that approaches and approaches the minimum density D _min and the maximum density D _{max as} long as it is a non-linear function. The inclination can be changed around the point symmetrical point of f _i (X−b _i , a _i ), which has the effect of making it easy for the operator to adjust the gradation curve.

According to the above (4), a gradation curve can be created based on pair information of (pixel value, density) associating a pixel value with an output value (density or luminance) in one-to-one correspondence. There is an effect that it is possible to realize gradation processing that cannot be handled by a gradation curve prepared in advance.

According to the above (5), increasing the natural number N in Equation 1 means increasing the number of fitting parameters. Since the degree of freedom increases as the number of parameters increases, there is an effect that various gradation curves can be created. Also, increasing the natural number N sequentially from 1 is as follows.
There is an effect that the number of parameters for obtaining a desired gradation curve is reduced as much as possible, the calculation time and the number of input information (pixel value, density) are reduced, and the operator can easily make adjustments.

According to the above (6), when the number of pairs of (pixel value, density) information is small, especially when there are two points, fitting can be performed by parallel movement with the gradient of the gradation curve.
There is an effect that a gradation curve having a complicated shape can be easily adjusted.

According to the above (7), by reading the density (or luminance or the like) from an image photographed by a certain photographing device, gradation processing similar to that of the photographing device can be performed, and two different kinds of photographing devices can be used. There is an effect that images can be compared.

In the adjustment of the gradation curve, it is often the case that the parameters are adjusted while looking at the image and then the image is viewed. According to the above (8), the image and the adjusted gradation curve are displayed side by side. This has the effect of making it easier for the operator to make adjustments.

According to the above (9), it is possible to grasp which point of the gradation curve to be adjusted corresponds to the designated point of the image, so that the operator can easily adjust the point.

According to the above (10), the operator can change the gradient of the gradation curve only in a certain region of interest, and there is an effect that a more desired gradation curve can be obtained.

According to the above (11), the operator can change the gradient of the gradation curve without changing the density of the region of interest in the image, which has the effect of making it easy for the operator to adjust.

The present embodiment can be realized by a computer executing a program. Also,
Means for supplying the program to the computer, for example, a recording medium such as a CD-ROM in which the program is recorded, or a transmission medium such as the Internet for transmitting the program can also be applied as an embodiment of the present invention. The above-described program, recording medium, and transmission medium are included in the scope of the present invention. As the recording medium, for example, a flexible disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a magnetic tape, a nonvolatile memory card, a ROM, and the like can be used.

It should be noted that each of the above-described embodiments is merely an example of the embodiment for carrying out the present invention, and the technical scope of the present invention should not be interpreted in a limited manner. It is. That is, the present invention can be implemented in various forms without departing from the technical idea or the main features.

[0161]

As described above, according to the present invention, the gradation curve is formed as a linear sum of the functions f _i (X−b _i , a _i ), so that gradations having different inclinations in different pixel value regions are obtained. A curve can be created, and a gradation curve can be created according to various images.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an embodiment.

FIG. 2 is a flowchart illustrating a processing procedure according to the embodiment;

FIG. 3 is a diagram illustrating an example of a subject image for acquiring pair information of (pixel value, density) associating pixel values with output values (such as density or luminance) in a one-to-one correspondence;

FIG. 4 is a diagram in which obtained (pixel value, density) pair information is plotted on a plane.

FIG. 5 is a diagram illustrating an example in which a gradation curve is fitted based on (pixel value, density) pair information.

FIG. 6 is a graph showing characteristics of the functions shown in Table 1.

FIG. 7 is a graph showing characteristics when the parameters a _i of the functions in Table 1 are changed.

FIG. 8 is a graph showing the characteristics of the linear sum of the functions in Table 1.

FIG. 9 is a graph showing an example of fitting when the pair information of (pixel value, density) is one point.

FIG. 10 is a graph showing an example of fitting in a case where the pair information of (pixel value, density) is two points.

FIG. 11 is an LUT showing a relationship between an input pixel value and an output pixel value.
It is a figure showing the derivation procedure.

FIG. 12 is a graph showing an example of a method for acquiring pair information of (pixel value, density).

FIG. 13 is a graph showing an example of a method of acquiring pair information of (pixel value, density).

FIG. 14 is a graph illustrating an example of a method of acquiring (pixel value, density) pair information.

FIG. 15 is a graph showing a method of changing a gradient of a gradation curve around a fixed density point.

FIG. 16 is a graph showing characteristics of a gradation curve of a conventional example.

FIG. 17 is a graph showing characteristics of a gradation curve of a conventional example.

[Explanation of symbols]

 110 source 120 subject 130 sensor 140 A / D 150 operation unit 160 image processing unit 170 image display unit 200 image information acquisition unit 210 gradation curve derivation unit 220 error determination unit 230 gradation processing unit 240 parameter storage unit

Continued on the front page F term (reference) 4C093 AA26 CA15 CA21 FF08 FF21 5B057 AA08 BA03 CA02 CA08 CA12 CA16 CB02 CB08 CB12 CB16 CC01 CE11 CH08 5C077 LL04 NN02 PP15 PP47 PQ12 PQ20 SS06

Claims (13)

[Claims] 1. A relationship between a pixel value and an output value corresponding to the pixel value.
Image information acquiring means for acquiring image information to be represented, and tone curve deriving means for deriving a tone curve from the image information
The tone curve deriving means has a function F_i(Xb_i, A_i)of
A linear sum, C_i, A _i, B_i, D_min, D_maxParame
Use the model function expressed by the following equation
An image processing apparatus characterized by the following. (Equation 1) 2. The function F _i (X−b _i , a _i ) is a monotonically increasing function except for the case of a straight line, has the properties of the following equations 3 and 4, and furthermore, the coefficient C _i of the equation 1 is 2. The image processing apparatus according to claim 1, wherein the following two conditions are satisfied. (Equation 2) 3. The function F _i (X−b _i , a _i ) is one of three functions shown in Table 1 below or a combination thereof. The image processing apparatus according to any one of the preceding claims. [Table 1] 4. The gradation curve deriving means, based on pair information of (pixel value, output value) associating the pixel value with the output value in a one-to-one correspondence, a function F _i (X−b _i , a _i) A tone curve deriving unit that derives a tone curve to be fitted using C _i , a _i , b _i , D _min , and D _max of Equation 1 as parameters using a model function represented by Equation 1 which is a linear sum of Equation (1). 2. The image processing apparatus according to claim 1, comprising: 5. An error discriminator for discriminating an error between pair information of a pixel value and an output value in a one-to-one correspondence (pixel value, output value) derived by the gradation curve deriving means. 5. The gradation curve deriving means according to claim 4, wherein the gradation curve deriving means increases the value of the natural number N in Expression 1 if the error is equal to or greater than a certain threshold, and derives the gradation curve again. Image processing device. 6. The tone curve deriving means includes a parameter
C_i, A_i, B_i, D_{m in}, D_maxAnd the derived expression
Substituting (X−B) / A for the gradation curve Y (X) of 1
((X-B) / A)
A and B are introduced, and the pixel value and output value selected on the image are
Pair information of one-to-one correspondence (pixel value, output value)
Adjust the parameters A and B to fit
The image processing apparatus according to claim 4, wherein: 7. The pair information of (pixel value, output value) associating a pixel value and an output value in a one-to-one correspondence for fitting the gradation curve is obtained by photographing a subject whose output value changes stepwise. The image processing apparatus according to claim 4, wherein the acquisition is performed by inputting an output value to the pixel value of the corresponding area of the subject image. 8. The image processing apparatus according to claim 4, wherein the fitting result of the gradation curve can be confirmed by displaying both a gradation processed image and a gradation curve. 9. The image processing apparatus according to claim 8, wherein designated points (pixel values, output values) are plotted on the image displayed on the gradation curve. 10. The pair information of (pixel value, output value) associating a pixel value and an output value in a one-to-one correspondence for fitting the gradation curve is registered for each gradation curve for which a parameter is determined. The registered (pixel value, output value) point is plotted on the gradation curve. By moving the plot point, the value of the (pixel value, output value) point is changed. 9. The image processing apparatus according to claim 8, wherein a parameter of the gradation curve is fitted to new (pixel value, output value) pair information. 11. The method according to claim 1, wherein the gradation curve is a parameter b
by substituting b _i -Dp to _i, by introducing a new parameter Dp and Formula 10 below, by changing the value of a parameter Dp of formula 10, fixed output that is central to vary the inclination of the gradation curve 2. The image processing apparatus according to claim 1, wherein an output value of the value point is changed. (Equation 3) 12. The image processing apparatus according to claim 1, wherein said output value is density or luminance. 13. An image information acquiring step for acquiring image information representing a relationship between a pixel value and an output value corresponding thereto, and a linear sum of functions F _i (X−b _i , a _i ), wherein C _i , a _i ,
a gradation curve deriving step of deriving a gradation curve from the image information using a model function expressed by the following equation 1 with b _i , D _min , and D _max as parameters. . (Equation 4)